128: Across the Universe (with Natan Last)

0:00We don't have a ritual. At the end of every show, we go pew, pew, pew. But at the beginning of the show, we don't really have anything. Gird your loins for linguistics. I think we could workshop that. Put that one on the maybe pile, shall we? Mm-hmm.

0:30Hello, and welcome to Because Language, a show about linguistics, the science of language. I'm Daniel Midgley. Let's meet the team. First, it's true stalwart, Hedwig Hirgaard. Hi, Hedwig. True stalwart. Stalwart is to be, like, irresistant, to be strong. Is that true? That's right. And you're here. You're always here. You're getting up sometimes very early in the morning. I'm not always here.

1:00I've got word games on the mind. What's your favorite word game that you play? Do you have a favorite daily word game or daily other game? I don't play every day, but I was playing Contexto recently, which is quite fun. That's the, like, user-friendly version of Symantle, right? Yeah. Yeah.

1:22Symantle's pretty tricky. You've got to... I don't actually... Contexto isn't any easier, is it? It's just presented in a way that is, like, nicer and more user-friendly? Or is Contexto easier? I remember Symantle being pretty hard. Symantle context. It's hard. It invites you to see the similarity between Mandible and Toaster. You're right. That is very similar. I get it. I get it. There's, you know... Both use word embeddings. I think you're right. I think they're essentially the same thing.

1:52Secondly, it's mostly true, but some parts may have been embellished, stalwart. Ben Ainslie. Hey, Ben. Same question. I... I... You didn't ask a question, I don't believe. We just riffed. A favorite word game. We did a Robin Williams. Oh, favorite word game. Sorry, sorry, sorry, sorry, sorry. It was that long that Ben actually forgot the question. That's not hard. It could have been a three-second answer and I would have done that. Um, I did enjoy the mini crossword before the evil empire that is the New York Times paywalled it.

2:25Connections, fun. I don't mind connections. Wordle, I actually don't love. Like, I'll do it, but it's quite far down my list. I actually hew towards the non-linguistic riffs on those games. So, I really like Framed, the movie guessing game with the frames, the still images. Uh, I really like the various geography ones as well. So, world all and global are both games I like very much. Very cool. My favorite right now is Jiznep.

2:57It's a... It does a rather long quote and it's arranged in five or six different lines because it wraps around. So, there they are. All the... Oh, oh, shit. Okay. So, it's like six lines deep and at the top of each column is all the letters you'll need for that column. Oh, okay. So, it's a bit Sudoku-y a little bit, right? Maybe a little bit. Oh, yeah, a little bit. You know what? For the purposes of our discussion, I'm going to say yes. Okay, good. Word games are on my mind because across my desk came a book, Across the Universe, by Natan Last.

3:29Now, if you're not in the cross world, you might not know who that is. But if you are, he's a legend. He actually got a crossword puzzle into the New York Times at age 16, the youngest person to do that at the time. Wow. And he now does crosswords, but also poetry and asylum seeker advocacy. He does all of his stuff in the pages of the New Yorker and the New York Times. And the book, Across the Universe, is a tour through the history of crosswords and the future of crosswords. When I get a book like this, I try to skim it for the interview, and I couldn't skim it because I just, yum, yum, yum, yum, yum, yum, yum, yum.

4:07It was too yummy. So, I'll be talking with Natan Last in the middle part of our show. That is really exciting. Very nice. Yeah. We want to give a big shout out to our patrons. They keep the show going with their words, news, ideas, related or nots. But they're also just kind of hanging out and being interesting on our Discord. So, we would like you to come and join in. You'll be helping the show, keeping us going. Even if you're $1 a month or $100 a month, we'd like you just the same. Or even be a free patron.

4:37Come see us. We're patreon.com slash becauselangpod. Hey, Daniel. Mm-hmm. I got a fever. Fever for what, Ben? I got a terrible fever for linguistic news. And the only cure is you dropping some hot information in my lap. What have you got? Okay. This is just me or is that just like become so sexual? Like, especially like I have a fever. Hot news. Ugh. Yeah. These metaphors, right? Metaphors of hotness.

5:08We're in a sexy space. I knew what I was doing. You know what, Ben, though? There's nothing that you can do about this fever. There is no cure because when I try to give you the linguistic skinny, the only cure is more. More. Oh, wow. Yeah. Look at him go. You really liked crude limericks when you were a kid, didn't you? I can tell. I'll have to ask the man from Nantucket. Anyway, this one is a new one from Dr. Adele Goldberg of Princeton University and Dr. Shahar Schertz, published in Language. Remember Hedwig talking to Adele Goldberg in our episode 121?

5:41I do. I'm so good. Well, this is the latest work from Adele, so, you know, of course, had to devour it. Ben, you were kind of surprised by us saying that linguists don't really know what a word is. Could you just elaborate on that a little bit? Because I thought that was... I think I might have gone on a registered trademark Ben mini rant on this when it came up last time. Kinda. It is always amusing to me when you find out that a field of knowledge, once you dig deep

6:16enough and get, like, in the weeds enough, actually has some serious issues with what an outsider would assume is just, like, baseline agreed knowledge in that field. For instance, you mentioned that linguists don't really have a cohesive agreement on, like, what a word is, which to me is a bit like a doctor, or doctors generally, not really having an agreement on what, like, cancer is, really.

6:47And now, I don't know this for sure. That might actually be true, for all I know. The other day, someone was like, cancer is actually, like, a rather broad concept, and, like, maybe endometriosis is cancer. Yes. Yeah. Like, why don't we call endometriosis cancer? And, yeah. Or, like, certain kinds of endometriosis when they start producing their own estrogen. And it's like, that's what a cancer cell does. That's a cancer. Anyway, so, I think this happens in a lot of fields. You get a lot of smart people in a room. There's no reason for anyone to be defensive here. It's just as lay people.

7:18No, no, no. I'm just, I'm just, I'm just, I want to turn the tables, Ben. Hey, Ben, tell it out, Ben. Don't just do it too, Ben. Come on. Shut up. Well, this paper is called The English Phrase as Lemma Construction. When a phrase masquerades as a word, people play along. Phrase as lemma. Now, when this says lemma, just think word. That'll, that'll do us for now. So, for example, Ben, when I say that something is a once-in-a-lifetime opportunity. Mm-hmm. What is once-in-a-lifetime?

7:50What part of speech would you say that is? A verb, a noun? Once-in-a-lifetime, I would probably have said is maybe an idiom? Well, yes, it is. Yes, it is. Yes. Would you say that it's doing anything noun-y or verb-y or something else? Uh, it is. Wait, wait, wait. It's probably noun-y, right? Because it is describing the- no, sorry, it's, it's, um, what's a describing word? Uh, um, adjective. It's adjective-y because it's describing the nature of something else, right? Okay. Comes, it's doing the describing, comes before a noun, right?

8:22Looks kind of adjective-y, kind of? No, I think? Um, well, it could also just be another noun because we have compounds nouns in English. You can say, like, uh, a horse-gull is, horse is not an adjective. A leather jacket. A leather jacket, et cetera. So, a once-in-a-lifetime opportunity. I also would say once-in-a-lifetime, I'm not even, I'm not sure it qualifies as an idiom because you can take it apart by its parts, and it means what it means. That's true. Well, we have talked about lots and lots of phrases like this that sort of masquerade as

8:53words. For example, a marry-me chicken dish. We've talked about both-sides-ism. We've talked about a lot of these. So, this paper describes these as a phrase as a lemma. It's a phrase that behaves as if it were a simple word. So, here's one of the contentions of the paper. They argue that we can't shoehorn them into other categories like nouns and verbs because they don't have the same distribution as adjectives, for instance. They don't act the same as adjectives. They use an example.

9:25Here we go. Stephen Colbert, in a show, there was a story about how Selma Hayek's owl coughed a rat hair ball on Harry Styles. I guess that was the thing that happened. Told the story on Stephen Colbert's show. And Colbert said, meanwhile, in Selma Hayek's owl coughed a rat hair ball on Harry Styles news. So, he was able to do the joke better than I could. But the paper says he treats the event of the actress's owl coughing up rat hair balls

9:55on the singer-songwriter as if it were familiar enough to be a lemma named by a word. It's like, when I do it that way, instead of saying it as a sentence, I'm using it as a phrase, as a lemma, and then I'm inviting you to pretend that this is common enough to be a thing. I'm making it a thing. I'm making fetch happen. Well, you're talking to your audience who listened to that story at some point. So, you know that they heard it.

10:26You're not making fetch happen. You know that fetch is already in the waters. I wonder, though, I reckon you could do it the other way around. I reckon if you're a good performer, right? I think a person like Stephen Colbert, who I have a lot of performance respect for, could probably lead the story. Like, he could pivot from one story to another and turn to a different camera and be like, meanwhile, in Selma Hayek's owl coughs a rat hair ball on Harry Styles news, and then just basically repeats that. Selma Hayek's owl coughed a hair ball on Harry Styles.

10:56I think that'd be more a Norm Macdonald's thing, wouldn't it? Yeah, possibly. This brings me back to the fact that, like, if I had endless money funding-wise and got to do anything I wanted, I think comedy is under-researched because they're doing funny things with common ground, like setting things up and turning it around. Comedy is so tough as well. Like, there's so much to analyze. There's so much to analyze. And also, but, I mean, we kind of know what they do. I don't know how much you can, like, I would like to write something about something that's

11:29new. And I don't know if saying comedians play with words in a funny way is super new, maybe. But it's so cool. I want to know how they do it. And I also think they don't know. Let me get to some of the findings of this paper. First of all, when you pick up a phrase and use it as a lemma, they say they're treated as known, as a known entity, even if they're not, which makes them kind of funny. So if I say, oh, I can see that we're up to the cleaning the fridge stage of work avoidance, now I'm treating that like it's a known thing. And it's funny. It works.

12:00Yeah, 100%. It presumes that we share some common knowledge together, which is another thing that we love to do with language. Oh, no, you and I share a thing, right? And in fact, they even did some work to show that people judge these high-frequency phrases as lemmas, phrases as words, as implying more common knowledge, as being wittier, and being more sarcastic than if you had just gone ahead and used the phrase or the sentence in the regular way. They also mention British drunkenims. Now, we've talked about drunkenims maybe before on the show.

12:33The thing where you say, like, as long as you say any sort of noun, so you can say any sort of noun as long as you preface it by saying, I was totally. So look around you and name an object in your house right now. I was totally video game controlled. I was spaghettied. Well, that's an actual word. I was totally spaghettied, I swear to Zeus. I was totally scarfed. I was totally... Totally scarfed, right? Yeah, yeah, I get it. I was totally mattressed. And for some reason, it doesn't matter what the noun is, that construction is already set

13:05up in the I was totally part, and then you could just go ahead and do it. So I think that this paper says a lot about the constructionist view, the idea that we have constructions in language, these bits of our repertoire that we can sort of pick up and use. These phrases as lemmas are constructions that we can just go ahead and use in different ways. They're not nouns. They're not verbs. They're not adjectives. They're phrases as lemmas. They act differently. And this is an interesting look at how we understand words and how we understand what someone's

13:38doing when they do what we don't expect. I feel like English just is a lot. I wonder how true it is for other languages. But... They do give some non-English data as well. So we'll slap up a link up to the paper on our show notes for this episode because language.com. And I noticed as well that it looks like the PDF is available for free, which is very... Yay! Yay for open access. Everyone's favorite. Now, have you ever thought that there was an emoji that existed and then you found out that it didn't? Oh, this is like a Mandela effect for emojis.

14:10Yes, it is. Mine's binoculars. But I have also found that some user interfaces will only show me some emojis also. Really? Yeah. Well, certain... Like, it's funny, right? Like, so I only noticed the other day that Outlook has its own suite of Microsoft-specific emojis. That too. When you put it into... So, which then makes me wonder, okay, maybe they haven't replicated all of them. So maybe that's something that you're encountering is they have like a reduced library or something.

14:41Well, sometimes if you're like on Discord on the phone, for example, and you're typing a message and you click on the Discord symbol for emojis and you look through those emojis, you get a slightly... You both get the like Discord server stuff, but even for the regular emojis, I don't think it's the same as when you go to your keyboard and you click the emoji. Mm-hmm. Um, I don't... I've never experienced this because I have used... I would describe myself as a, quote, basic bitch when it comes to emojis. I use the like, the six or seven that I think we would describe as like broadly pan-universal,

15:16right? Like the smiley, the bigger smiley, the flat mouth, and like maybe the kissy with a wink and hearts or something like that. And that's it. Okay. Okay. So they all exist as far as I'm aware. They still do, yes. They're very common. Now, has anyone heard of the seahorse emoji? Do you remember back in your mind seeing a seahorse emoji back there? Maybe. There's a lot of, there's a lot of animals. There's like a... I find as well, I don't know if you experienced this, Daniel, but here in Australia, for some

15:47quite bizarre reasons, seahorses... I see seahorse iconography everywhere. So I'm also aware of the fact that I might just be projecting many, many, many, many, many seahorses that I've seen around the place into an emoji form. But having said that, I feel like I have seen a seahorse emoji. Okay. Well, you haven't because there isn't one not in the standard Unicode set. Biscuits. However, there's a funny story from Josh Milton on metro.co.uk. It's called, we asked chat GPT if there's a seahorse emoji.

16:17Things got a little weird. So I tried it myself. I went to chat GPT and I said, I think I remember a seahorse emoji. Is there one? Can you see my screen right now? Yes. And it's a seashell. Yes. Seashell. And you're right. There is a seahorse emoji. Fish. No, wait. That's a tropical fish. Here's the actual one. Dragon. Nope. Wrong again. Aha. Emoji. This one. And check out what's going on. It does like 17 iterations all the way down the page of saying, oh no.

16:48Okay. I got this. The real one is a string of garbage emoji that aren't relevant. No, wait. Okay. I'll stop joking. Yes, there really is one. No. I have never encountered... I've never encountered doing this kind of... No. I have encountered it saying, hold on, I got to think for a minute. No, and what I've also, like I, one of the principal criticisms I've had with most large language models is that it would stop at the front, in my experience. It would say, yes, and it would give you a seashell and it would basically be like, I

17:21did a good job.

17:24It's been very comfortable being wrong in all of the things that I've found it be wrong in. What we wanted to do is do all of that thinking, reasoning, whatever thing that is visualizing that it's doing. And then at the end say, I'm sorry, there isn't one. But we don't want to... Why want to see this? Yeah. So have we seen an instance of like behind... Are we seeing behind the curtain here? Is that what's going on? I think we might be. And I think this has interesting implications for how large language models process text. So I think there's a combination of things happening here.

17:56First of all, the large language model, and by the way, Claude does the same kind of thing, although it didn't thrash for 17 iterations. It did maybe three and then stopped. But it's got access to text where people are convinced that the seahorse emoji exists. People are talking about the seahorse emoji as if it exists because of the Mandela effect. So that primes the large language model to say, this must be a thing. All these people are saying it's a thing. But that can't be the whole story because I tested it on other Mandela effects,

18:26like the Fruit of the Loom logo, which many people are... Can you describe the Fruit of the Loom logo, the underwear brand in the USA? I have no idea what it is. It's a sort of shell from Greek mythology overflowing with fruit. The cornucopia, full of fruit. Cornucopia. That's the Fruit of the Loom logo, except it's not. The Fruit of the Loom logo has never had a cornucopia. It's never had a horn of plenty thing with all the stuff coming out. But many people are convinced that it once did and it never did.

18:57But when I ask ChadGPT and Claude about this, they both say, oh no, that's a straightforward Mandela effect. That's not how it was. Interesting. But is that because there's not enough information about the seahorse effect? Sorry, the seahorse emoji. Maybe because there aren't enough websites that talk about, no, this is not a thing. Because there are a lot of Mandela emoji. The binoculars, the vacuum. Also, I tested it on the Monopoly guy. What's the Monopoly guy look like? You know, the game Monopoly.

19:28He's an old guy with a... Hat and a monocle and a mustache. He's got a hat, he's got a monocle and a mustache, except one of those three things he doesn't have. Yeah, he doesn't have a monocle. He doesn't have a monocle. And if I ask ChatGPT, am I freaking anybody else out here? Okay, fine. Well, I just asked my ChatGPT about the seahorse, and it went on, I had to, it never stopped.

19:53And I'm sorry, but then I asked something else I know is untrue, and it gave an answer, it stopped. Just gave an answer. That's what happened with the Monopoly guy. It's almost as if there's like a... You know how Google hardcoded in, like, what's the meaning of life 42? Right, right, right. It almost feels like someone has... What I'm wondering is if this is in some way related to... Like, I've noticed that all of the things that you said, Daniel, are in some way visual.

20:23Yes. Does it look like this? Does it, like, all of them, right? Like, if you ask ChatGPT, how do you spell the Berenstain Bears, right? One of the more famous Mandela effects. I imagine it's going to spit out either the correct answer or the wrong answer very, very quickly. Are we encountering an issue with a large language model fundamentally not having any visual acuity whatsoever? So all it can do is approximate based on what it has found in the corpuses of text that it has been trained on.

20:55Yep. The fish emoji doesn't look like a fish emoji to a large language model. It looks like a space in the Unicode directory. And also, those emoji are scattered around in discussions, but the emoji being used has sometimes a tenuous connection to the actual words themselves because they are functioning as tone indicators or sometimes gestures. Their meaning is a bit obscure. So I think the fact that this is about emoji is contributing to the confusion. It thinks that it can pull something out

21:27because people talk about it as if it's real. But then when it tries to do so, it spits out weird random things and it doesn't, it can't see them. It can't see the little pictures. No. Yeah, yeah, yeah. And the text that it's trained on doesn't give much clue about what it is. So I think this is a fascinating little insight about large language models and emoji. Neek eye. And I think you're right. There's something specific about emoji, not just other visual cues. Because I just asked it like, is the Doctor Who telephone box red? And there's lots of people in fan fiction and whatever

21:59writing about it and describing it as blue. So it knows that it's blue. Like it knows to associate those words. So it's something very particular to emoji. Is it blue? That's cool. It is blue. It's blue. Okay, sure. And now it's time for Related or Not. Which one have we got this time? Okay. I've got. I don't know. What do you want? I don't know. We need some fresh ones. We need to put the call out. We need some sick new. I want to, here's what I want. I want a Jungle Related or Not intro song.

22:33If someone out there can bring me like a real like Related or Not, like that kind of thing. We're leaning away from AI though, because it's stolen data. Sure. No, I'm asking people to go and make it on like some kind of like little like free loops or something. We'll play it. Yeah, we'll play it. For sure. Well, okay. Well, this one comes to us from Duck Off of Water's Back on our Discord. Great name. Pharaoh Cat on our Discord. And also James on our Discord. You know, it's all happening around here. These are three pairs.

23:03Two of them are related. One of them's not. You got to pick the distraction. Here we go. Okay. Anguish and languish.

23:14Anguish.

23:17Languish. So, anguish is to be like, to have anxiety and languish is to like, be lazy, languor about, move slowly. I feel like languish, yes, there's a bit of lassitude involved in languishing, but you could also be, it could be bad. It could be like, I was, if you're languishing, it's never good. Yeah, it's definitely a negative valence. I've always thought of it like, in relation to boats, for some reason, like sailboats would languish if there was no wind and that sort of thing.

23:50And then anguish, which is a feeling of desperation and sadness. Okay. That's one. Second, experience and expertise. And for this one, because clearly the X is related, these are both Latinate, I want to focus on the per or the pert. That's, that's the part I'm interested in for this one. Yeah. Okay. And third, from James, camaraderie and camera, camaraderie, which is friendship and well-being and camera, the little clicky device that we have on our phones, but a real one.

24:25Okay. So those words again, here are the three pairs, anguish and languish, experience and expertise, camaraderie and camera. Two of them are related to each other, but the third one's a distraction. Which one? Oh, I've got a gut feel on this one. I reckon anguish and languish are related. It just seems too, too, too French for it to not just to be la anguish. Oh, I see what you mean. And then some English people being like, they spelt it languish.

24:56So like one came over at one stage and then another one came over at another stage. I actually think our little ex-pals are the false friends here. What was it? Experience and... Expertise. Expertise. Yeah, I reckon they're not friends because I have a feeling, again with French, that when the Renaissance men and women of the early days of photographic technology were doing that sort of stuff, the word they used for the device could in some way be related to

25:30like friend or friendly or friendship or something like that. Mm-hmm. Okay, that's my... So the exes are the sneaky devil. I feel like per and pert, I could see how those would be different somehow. Yep. Okay, so you're going for experience and expertise. Hedvig, which one are you going for? This was a really hard one. I think you pulled together really fun examples. So I'm also on the anguish and languish are related train, but I think it's one of those pairs like willy-nilly, like they're actually opposite pairs or used to be at some point.

26:04Yeah, okay. And then the thing about expertise and experience is that because you hear the ex, but if you look at the periods and perties, the thing is quite small that's similar actually, and it could be chance. The thing about camaraderie and camera, I mean, so for an English camaraderie camera, the first vowel is different. So I feel like it's some sort of loan.

26:35Is camaraderie, is comrade a loan from Germanic or Slavik? French? I feel like it's a... Rottery, yes. But camera, it's to do with shutters, right? It's to do with closing. Oh, interesting. I'm not sure. I don't actually know the etymology. It's to do with a room. I thought it was room. Yeah. It's room. It's a little tiny room with a shutter. Yes. It opens. Oh my God, I'm having... You know when you have a thought and you articulate at the same time as you're having it?

27:06You're having that feeling like when you know two different parts of the city and then you find a road that joins them and you're like, oh, I didn't know these two parts go together. Oh my God. Yeah, yeah, yeah. Or when you're in a city and you'd be taking the subway all the time, the metro, and then you realize that places are connected above ground.

27:28There is an up world. There's an up world. Yeah. Okay. It's really hard. Experience, expertise or comrade. Camaraderie, camera. I think camaraderie and camera are the false pair. Okay. That's good. Different for me. I think this is interesting because we're all three different because I thought anguish and languish. There's going to be a winner. This is great. I know that in musical directions, if it says langsam, it means sad because that's German. Or is that slow? Yeah. No, it's sad. No, it means slow. Anguish.

27:59Yeah, that's it. And then I thought anguish is related to angst. I just thought, yeah, that peels apart into two different things that I know about, sort of. So, I'm saying those are the distraction. And the answer is, I was right. Anguish and languish are the distraction. The worst possible outcome. Latin angustia is tightness or difficulty. And then Latin languere is to be weak or to faint. Two different words. They've come together.

28:30English is just like into them together. That happens a lot. Like the different colored Play-Dos in the hands of a toddler. As far as experience and expertise, that pair or pert is they're both Latin peritus, meaning experienced or tested. It's a thing that you've been through. So, and Hevig, you are quite correct. Camera is a room. So, James says they are, in fact, related camaraderie comes from comrade, which comes from someone who sleeps in the same room or your chambermate.

29:01And then it goes back to Latin camera, a little vaulted room, which gives us the box that captures a moment in time, the device you might whip out to document moments of camaraderie. There we go. There we go. Wow. Jinx. Now, second one. People often misspell these two words. They spell loose and lose the same sometimes. Ah, yes. Related or not. Yeah. Loose and lose. Oh. Yep. Are they related? Loose and lose.

29:31Loose has two O's and lose has one. Tricky. I say not related. No. I say not. Well, I have to go related then. Screw you guys. And I'm going to justify it, even though I didn't have any reason to. I think semantically there's a connection there, right? When you loosen something, you lose. It's the quality that it had before, right? So, that's where I'm sitting with that. It's tenuous. If it gets loose, it might get away and then you lose it.

30:02Yeah. Basically. Yeah, I said no because I just thought, nah, those two words feel like super old Germanic and I can see how they would accidentally grow together, but be really different. All right. And? Up to Hedwig now. Oh, she said no. I also guess that they're not related simply because I could match one of them to a Germanic root and the other one not. And I thought English is like borrowing something or doing something strange. Okay. So, we have one vote for yes and two votes for no. Answered.

30:32No, they come back to totally different words. Old English, losian, is where we get lose from, to be lost or to perish. And then lus comes from Old Norse laus, among other places. I just thought it was interesting that they look so similar. I mean, lus has one o, lus has two. But if we look at the forms, if we click on the forms tab in the Oxford English Dictionary, we see the word lose written with two o's sometimes in the 1500s, 1600s, and we actually find loose written with one o in the 1500s.

31:03But now we pretty much settled on lose and loose, one o, two o's respectively. What a frustrating time to be alive. But Ben, I'll give you that. If this wasn't in the game setting and I wasn't also thinking about game mechanics, I don't know what I would have answered because they are really similar. And I think if there's anything we've proven with this segment, related or not, over many episodes, it's that you can work your way to a connection or to a not connection with a lot of these pairs.

31:34It's hard work. You can justify a crazy amount of bullshit. It is true. Yeah, you can justify a crazy amount of bullshit. And sometimes the semantic string that holds an actual reconstruction together appears very loose. And this is why... I didn't see it. And that's why you're loose. That was unintentional. I was just primed. But that is why historical linguists use more than isolated pairs, right? That's like why they don't do this. All right. Last one.

32:05This one's different. A different challenge that I wrote. I looked at the word upholstery. You know, when you reupholster a couch, it's the way that the leather is arranged on a couch or a chair. Upholstery. Which of these words is it related to? Okay. Is it holster? Like a gun. You stick a gun in a holster. Luster.

32:29Shiny and velvety. Pole. Pole. P-O-L-E. Or uphold. Is upholstery related to holster, luster, pole, or uphold? Okay. I won't guess on this one. I think I've got my answer on this. But I want to give Hedwig first right of refusal. Because if she picks one, I've got to pick another one. There's no fun if we're both right or both wrong. It might be fun. I mean, why on earth wouldn't it be uphold? That's what I... Oh, damn it. Like, it has to be.

33:00I'm sorry. Why does it have to be? What's the relationship? That's exactly where I went. Because upholstery is doing that. It's upholding the fabric to whatever structure you are upholstering. Okay. That's my thinking there is, like, you take fabric and, like, foam and whatever else, which is notoriously very loose and flowy and not very good at staying on things. And then you use all sorts of wizardry, like glues and tacks and stuff, to keep it in a particular shape for many, many years. You uphold it.

33:31Yeah. That's where I... Yeah, that jumps straight out at me. But I can't choose that. So I'm going to go with pole. Okay.

33:40That was... The most distractee of the distractor answers that I could come up with. But okay. Okay. You didn't like holster. It was... That was the second most relevant to me in the list. The only reason I didn't think that it would be that is... I don't know if this is, like, a D&D thing. Maybe you thought this as well, Hedvig. Given my, like, fantasy knowledge and just general sort of, like, interest in, like, the ye olde days, leather workers and upholsterers were very different things and very different

34:15trades. And holsters were almost universally made of leather. Yeah. Being a leather worker was a very, um... I grew up with a leather couch. Okay. Fair. Fair. I think that's much more recent, though. And, like, upholstery is a concept with fabrics. I don't know. Anyway. And that's just what I thought. Okay. Here's the answer. It is uphold. You are upholding, keeping something from falling into disrepair. Sorry.

34:45I mean, it's what you wanted to guess. It's what you wanted. That's right. So, the relevant sense of uphold for upholstery, the way that they're related is that if you kept something from falling into disrepair, you kept it up. You were upholding it. You were an upholdster. Right. Stir meaning someone who does a thing. However, if you take a look at Edom Online, Douglas Harper also mentions that it might be a slightly different sense. An upholdster might be not somebody who upholds furniture and keeps it from falling into disrepair,

35:18but rather somebody who undertakes or carries on a business. You are upholding some kind of enterprise. Wouldn't that then be a word for every sort of business? It could have been at first, but then it focused on that one in the same way that Undertaker did. You know, it used to be that Undertaker was somebody who undertook a business, but because of taboos about death, we didn't want to say this person is a person who buries people. We would say, well, this person, you know, engages in undertakings.

35:49So this would be like if we, in a hundred years, only use the word entrepreneur for people who bury dead people. Wait, we do this. Hey, wait, wait, wait, stop. Entrepreneur is not any business. Isn't it? I'm lost. Entrepreneur is like specifically annoying startup people who are working, who are trying to do something new and disrupt business. Like that's what an entrepreneur is. She's right, you know. Or it's a hardworking immigrant.

36:19It's one of those two. No, you know what? I think we are applying some very contemporary like judgments to a word that didn't mean that. No, I'm into this though. No, but I'm talking about what it means now. These are the meanings that have attached themselves to the word now. Yes. Annoying startup guy or hardworking immigrant. Hardworking immigrant who sells dinner. Yeah. That's quite interesting. And a gig worker is people who are biking for Uber Eats. Okay. The gig economy. That's a gig. Right. Or a band.

36:50Yeah, I was about to say, like they've taken that from the music world. What a great word. Annoying startup guy, an immigrant hard worker, or someone who is biking to deliver food or playing music. These are really funny pairs. What's interesting as well is I don't think the people who are engaging in the gig economy describe the discrete units of what they do as gigs, whereas a musician does, right? They will talk about the gig I have on Saturday or whatever, but I don't think the Uber drivers say, oh, I've got a gig from the airport or whatever.

37:22They just say, no. No, but they would say, my gig is Uber Eats. Maybe. But then musicians don't say, my gig is playing music. No, so they're referring to it slightly differently. Yeah. I've got a gig tonight. Anyway, anyway, anyway. Well, fascinating stuff. Thanks to everybody for those words. And if you have a related or not pair you'd like to give us, go ahead and do that in all the regular ways. Hello at becauselanguage.com or on our socials. I'm here with Natan Last, who contributes crossword puzzles to The New Yorker and The

38:00New York Times, among others. He's formerly director at the Asylum Seeker Advocacy Project. At one time, the youngest person to get a puzzle accepted into The New York Times, which is pretty impressive. And he's the author of Across the Universe, The Past, Present, and Future of the Crossword Puzzle. It's Natan Last. Natan, thank you so much for coming and hanging out with me on the show today. Thanks for having me. It's an absolute pleasure. You have just come back from Crossword Con, which I've never been to. What was the vibe there? What would I have seen if I were there with you?

38:33Crossword Con is this awesome group of people that kind of showcase all the different ways crosswords tickle everyone's brain. So there's programmers showing the bespoke tools they've built to make puzzle making easier. I did a reading there of a more emotional section in the book, trying to talk about the ways that crosswords bring people together and even unite us in moments of grief. And there will be people there talking about crossword archive digitization, talking about the indie puzzle scene and the charity puzzle packs.

39:06It's just like a really great multi-layered cake cross section of the puzzle world. It's really wonderful. How do crosswords bring us together in grief? I have this chapter in the book about a pair of siblings whose mother, unfortunately, this was right at the start of COVID, was diagnosed with cancer. And they came back to New York where she was being treated. And even in the big hospital room that was in a new NYU unit, it had a big TV.

39:39They plugged in their laptop and they did the times puzzle every day. And a streak began and the streak kind of held or it seemed to hold the possibility of her illness doing away, the possibility of her life continuing. And it was just a way to mark the day, to collaborate. And for the mother who had virtually passed to see her children doing something together and to help. It's important to play together, isn't it? And this is one way for people to play. Yeah, exactly. It's traditionally an isolated pursuit, but now puzzles have become just like sharing

40:12our wordle bridge, a very social thing. That's a really good point. And the sociality of it, I mean, the digitization of crosswords makes it possible or easier to share. That's for sure. Because of something I read in your book, Across the Universe, I decided to try out a puzzle that you referenced. It was one by Matt Ginsberg. The title of it is Just Follow Directions, and it's from the 3rd of February, 2008. We'll have a link to it on the show notes for this episode.

40:43And I haven't experienced anything like that for quite a while. I have a habit of doing the mini in the New York Times with my young daughters, and they're just getting used to crosswords, and they're getting a vocabulary. It's a way for us to build vocabulary and a way for us to sort of work together to solve a problem, which is really cool. And it's cool that crosswords can do that. But this puzzle was just bananas. It's got words flying around everywhere. It's got ambiguous clues.

41:14It's got clever clues. The one that I think I noticed was punch. And I love clues like that, because you don't know if you're talking about punching someone or punching a ticket, like with punching a hole in something, or whether you're talking about a punch that you drink, and it's four letters, four letters long, and I've got to think, I've got to sort through all that ambiguity. So writing an ambiguous clue, I think, is part of the fun. But what else is fun about writing clues for you?

41:47Yeah, clues are this wonderful free associative game. I mean, the novelist George Parekh, who is a big language head and, of course, famously wrote a novel without the letter E in it, also wrote a weekly crossword for La Pointe and called the filling of the grid this letter-based arithmetic. It was like this tedious, maniacal task, where all that matters is that letter, words are this or that length. But filling a grid was a stroll in the thousand and one associations a word can evoke.

42:21And writing a good crossword clue is like that. At first, just letting your mind go to where it wants to go when it comes to what's evoked by a given word. But writing a really good clue, so there's the cultural reference point. There's what does this make me think of? And that first instinct is often the truest clue you'll write. But there's also clever clues. And that is honestly, if you've ever played the game Codenames, where you are looking at a word and you just are thinking of all of these kind of abstract nouns and verbs that

42:52are associated with that word. So I was recently helping a friend write a clue for the word make-out session, right? Make-out session. Okay, cool. And we wanted a clever question mark clue. And sometimes... A question mark. Hang on. For people not familiar with the syntax, that means it's a jokey answer. Is that right? Yeah, there's a pun involved. There's wordplay involved. And in this case, you know, like many previous crosswords, I wanted to pun on the fact that making out is sometimes referred to as necking, right?

43:25And so I just thought of as many two-word phrases as I could that had the word neck in them and eventually came up with neck stretch, which sounds like something your physical therapist recommends, but it's also a stretch of necking. Stretch of time. Exactly. And that kind of free associative hopping is, I think, what makes for really good, clever clues. The question mark, as you're saying, is often a way of hand-holding the solve for just a little bit and saying there's stretchy wordplay involved herein.

43:56Be careful. But also, you can, just like you're saying, use the bare syntax of crossword clues to hide proper nouns. So a clue might be something like law with many parts, which sounds like it's about some kind of ancient code of legal authority. But in fact, the answer is Jude, because Jude Law is an actor whose last name would be capitalized at the beginning of a clue. Wow. Okay. Yeah. It's that kind of groaners that really make us, really make us crossworders tick.

44:28Wow. Okay. And you know, the word neck, I don't know if anybody refers to kissing or making out as necking anymore. It's an old reference. And that was kind of an area I wanted to go to since we're talking about cluing. There are words that it's maybe time to retire. You know, like Nick and Nora's dog, Asta, which who knows that anymore? That's a hundred-year-old reference. You know, you think, maybe we better move on from that one. Are there any that you feel like it's time to retire?

44:59Or was there a clangor that you recently saw that made you go, oh, that one has passed? Yeah. I mean, there's a lot here. I think some of the, some of the like Hawaiian words for lava occasionally, or some of the name, the place names are places that I hadn't heard of. I kind of think many things are true here at once. You know, one is there's been a happy arms race to make crossword grids even cleaner. So even more devoid of this fusty crossword ease, as you're referring to it.

45:33And that's, that's definitely the term of, the term of art. These are words that appear so much more in the crossword grid than in real life. And in fact, there was this cool analysis by software engineer Noah Veltman that I'd quickly shout out, which is about the crosswordiness of a word. So crossword ease might be words that like era, era, year, area appear in crosswords so often. But of course, those words appear in real life too. So he kind of created this ratio that was how often does this word appear in the New

46:08York Times crossword grid against how often does it appear in the Google Books corpus, right? So words that are only in crosswords and the crosswordiest words are words like ursa. No one really calls the Big Dipper ursa major. Let's see what else. Smee, which is the captain, I believe from... Captain Hook's first mate. I was thinking of like NNE for North, Northeast, which I look at that and I just, oh my God. That's right. And you're pointing to a really good thing, which is not only is it frustrating as a solver

46:40to put that in, but the frustration begins when you see yet another clue. That's two cities that happen to be North, Northeast from each other in a pathway you'd never take as a traveler. And so it just feels so contrived. I would say those, I definitely agree. I'd like to retire. I think that there was this great project I talk about in my book called the Oreo Cluing Project, which recognized that Oreo, of course, is in every single crossword. And we can talk about, you know, the phonotactic reasons for that. There's a good reason that Oreo is in every crossword grid.